> ## Documentation Index
> Fetch the complete documentation index at: https://www.integrate.io/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# ETL: AI Transformation

> Use the AI Transformation to apply LLM-powered data processing to fields in your Integrate.io ETL pipeline. Supports custom prompts and models.

The AI Transformation component applies LLM-powered transformations to fields in your Integrate.io ETL pipeline. It operates per record (one request per configured field/record), supports prompt interpolation, and emits typed outputs you define.

## Overview

* Currently supports OpenAI Chat Completions models. Additional providers can be added on request.
* Accepts rows from an upstream component; outputs the responses to the prompts you configure.
* Prompts support `#{field_name}` interpolation from upstream columns.
* If `field_name` is not provided, it assumes your prompt is referring to the field for the row your prompt is on.
* Component alias is always unique.

## LLM API Key Management Options

Two approaches:

1. **Bring Your Own Key (BYO Key).** Supply your own OpenAI API key. All usage billed through your provider account.
2. **Integrate.io Managed Key.** Use an Integrate.io managed key. Generates a unique credential for your account. Usage processed through this managed key.

## Configuration

### Model Settings

* **Model selection**: Choose a Chat Completions model (e.g., gpt-4o, gpt-4o-mini)
* **Temperature**: Controls randomness (0.0--2.0)
* **Max tokens**: Upper bound on response tokens
* **API key**: Managed Key (provisioned via OpenRouter) or Bring Your Own Key

To view your provisioned key: **Settings > Your Settings > OpenAI API Key**.

### Select Fields & Preview

* **Select field name**: Pick the upstream column
* **Set prompt**: Write instruction. Use `#{field_name}` to interpolate values
* **Set alias**: Provide the output column name (unique by design)
* **Preview**: Click Preview to run on a small sample (5 rows by default)
* All AI prompt fields are returned as string datatype

## Testing

<Steps>
  <Step>Connect AI Transformation after a source or transform component.</Step>
  <Step>Configure fields, prompts, and alias.</Step>
  <Step>Click **Preview**; a small number of rows will be processed.</Step>
  <Step>Inspect outputs and iterate.</Step>
</Steps>

## Best Practices

* Keep prompts short and explicit; specify the desired output format.
* Use interpolation to add context from multiple fields.
* Use preview to validate behavior and manage token usage.
* You can use ChatGPT to help write prompts.

## Managed Key Costs

When using an Integrate.io Managed Key, AI Transformation usage is billed based on the LLM provider's consumption pricing. Charges are a direct pass-through of the provider's API usage. Billed monthly in arrears. You can configure a cost limit to cap spend.

Cost depends on: number of rows processed, input data length, output size, and model selected.